The Heritage Model

Shanea Watkins, at Heritage, writes:

On March 5, the House Budget Committee passed its fiscal year 2009 budget resolution. If implemented, it would increase taxes significantly, thus decreasing job growth, reducing disposable income, and weakening the economy. This WebMemo projects the likely impact of the House budget resolution on the tax burden, jobs, and economic growth in states and congressional districts.

The House budget resolution has the potential to cost the average American taxpayer more than $2,000 in additional taxes in 2012 alone. Higher taxes, particularly on capital, cause the level of private investment to fall, thereby slowing productivity improvements and weakening the earning capacity of households. Wages and business earnings, which are closely tied to productivity, would fall as well. As a result of the tax increases implicit in the House budget resolution, Americans could also see their personal income decrease by an average of $1,767 due to a weaker economy.

Moreover, the budget resolution could damage employment growth, causing about one million fewer jobs to be created, and could lower economic output by more than $100 billion compared to what it would have been; the average cost to congressional districts could be 2,191 lost jobs and $247 million in economic output.

There’s also some tables showing how much each state and congressional district is going to get messed up by these vile actions that Congress is taking. But wait… how do we know about these horrible consequences? (I’m assuming her caveat about “potential to cost” goes by the wayside. It seems to be glossed over quickly, but its pretty clear Watkins is certain this is going to hurt us bad.)

Analysts in the Center for Data Analysis (CDA) at The Heritage Foundation used two models to develop estimates of how the Bush tax cuts likely affect the U.S. economy. They used the CDA Microsimulation Tax Model and the U.S. Macroeconomic Model of Global Insight Inc., a leading economic consulting company. Estimates were projected to 2012 because that is the first full year when all of the tax provisions of the budget resolution will be in effect. These projections assume normal levels of economic, population, and employment growth over the next five years.

I crunch numbers and build models for a living. (Anyone need a consultant? Drop me a line!) As I type this, I’m running a simulation in a model I’m building for a client on another machine… I have about five minutes until it finishes its run and I can look at what it spits out and figure out where to go from here. (If you’re wondering, yes, much much of my writing takes place while waiting to see some results. Five minutes here, five minutes there, and voila, a post… or a chapter in my book.) So when I see mention of models, especially when those models spit out results that I’m not sure I buy, I’m interested.

Neither model is explained in any more detail in the rest of the post, so I went looking around to see what I can learn. If you google “CDA Microsimulation Tax Model” you find precisely two entries – one to Watkins’ post, and an ad in Job Openings for Economists where they’re looking for someone to maintain the model. Leave out the quotes and you get more hits – most don’t have anything to do with this model, and those that do come from other folks quoting Heritage, including such reputable organizations as Free Republic.

Global Insight is a reasonably well known consulting firm… and I found a bit more information searching online. Basically, it seems the model has about 1200 equations.. exactly what they are isn’t clear, since from what I can tell after a short search, the model is proprietary.

Now, this is policy that Heritage is peddling. Which means, they should try to be as transparent as possible, especially since they’re operating with a known bias. Which means that even if Global Insight’s model is proprietary, one would expect Heritage to make its model available. (Can anyone think of any reason why a policy think tank would keep its model proprietary?) At the very least, if I’m expected to accept the results of these models, I would expect to see the following:

1. A description of the results the model spit out in times past. For example, what happens when you pump in data available through the end of 1992, and the information about the 1993 tax hike? What about when you put in information available through the end of 2000, and information about the 2001 and 2002 and 2003 tax cuts? Does it tell you what happened with any degree of accuracy? If not, why not?

2. A description of the assumptions that went into the model.

I note that Watkins is very, very open about how results spit out by the model at the national level were allocated across states and Congressional districts, which is quite admirable. But it only makes me wonder that much more why we’re supposed to take the model itself, the model on which all her striking results are based, on faith.